Focused ion beam scan routine, dwell time and dose ... - CiteSeerX

IOP PUBLISHING

NANOTECHNOLOGY

Nanotechnology 18 (2007) 195305 (11pp)

doi:10.1088/0957-4484/18/19/195305

Focused ion beam scan routine, dwell time and dose optimizations for submicrometre period planar photonic crystal components and stamps in silicon Wico C L Hopman1 , Feridun Ay1 , Wenbin Hu1,3 , Vishwas J Gadgil1, Laurens Kuipers2 , Markus Pollnau1 and René M de Ridder1 MESA+Institute for Nanotechnology, University of Twente, PO Box 217, 7500 AE Enschede, The Netherlands 2 FOM Institute for Atomic and Molecular Physics (AMOLF), Kruislaan 407, 1098 SJ Amsterdam, The Netherlands 1

E-mail: [email protected] and [email protected]

Received 13 February 2007, in final form 21 March 2007 Published 17 April 2007 Online at stacks.iop.org/Nano/18/195305 Abstract Focused ion beam (FIB) milling is receiving increasing attention for nanostructuring in silicon (Si). These structures can for example be used for photonic crystal structures in a silicon-on-insulator (SOI) configuration or for moulds which can have various applications in combination with imprint technologies. However, FIB fabrication of submicrometre holes having perfectly vertical sidewalls is still challenging due to the redeposition effect in Si. In this study we show how the scan routine of the ion beam can be used as a sidewall optimization parameter. The experiments have been performed in Si and SOI. Furthermore, we show that sidewall angles as small as 1.5◦ are possible in Si membranes using a spiral scan method. We investigate the effect of the dose, loop number and dwell time on the sidewall angle, interhole milling and total milling depth by studying the milling of single and multiple holes into a crystal. We show that the sidewall angles can be as small as 5◦ in (bulk) Si and SOI when applying a larger dose. Finally, we found that a relatively large dwell time of 1 ms and a small loop number is favourable for obtaining vertical sidewalls. By comparing the results with those obtained by others, we conclude that the number of loops at a fixed dose per hole is the parameter that determines the sidewall angle and not the dwell time by itself. (Some figures in this article are in colour only in the electronic version)

possible route towards increasing integration density by several orders of magnitude is using so-called ‘photonic crystal slabs’ (PCSs), which have recently received much attention. By introducing ‘defects’ in a PCS in a controlled manner, one is able to implement passive and active building blocks (e.g. filters [2], sensors [3], and lasers [4]). Important breakthroughs have been reported on nano-cavities having Q -factors [5] exceeding 105 [6] and their applications as filters and even as all-optical transistors [7]. The reduction

1. Introduction The ultimate goal of integrated optics is to achieve full control of light propagation. Functional devices based on classical planar optical waveguiding in low to moderate refractive index contrast systems are reaching the limits of achievable complexity due to the necessarily large bend radii [1]. One 3 Present address: Fiber Optic Sensing R&D Center, Wuhan University of Technology, 430070, Wuhan, People’s Republic of China.

0957-4484/07/195305+11$30.00

1

© 2007 IOP Publishing Ltd Printed in the UK

Nanotechnology 18 (2007) 195305

W C L Hopman et al

Perfectly vertical sidewalls are, in principle, required for photonic applications to guarantee low-loss propagation; sidewall angles of 5◦ can already induce a 8 dB mm−1 propagation loss [33]. Fabricating submicron holes in silicon with perfectly vertical sidewalls by FIB milling is challenging due to the redeposition effect [23, 34–36]: sputtered Si atoms or clusters can be redeposited locally. This effect is less pronounced in the milling of slits (or trenches), which can result in sidewall angles up to a few degrees for low currents [37]. For similar structures as studied in this paper, sidewall angles of ∼9◦ have been reported obtained using a multi-pass type raster scan [38]. The redeposition effect becomes more dominant when the size of the holes is decreased. A straightforward solution for lowering the effects of redeposition by using a repetitive pass system was published in 1984 by Yamaguchi et al [39]. That work forms the basis for investigations on the milling properties that can be manipulated by varying the dwell time [40–42] (the time the beam is stationary at a fixed point), which may influence the number of repetition loops. However, dwell-times cannot be made arbitrarily small, because FIB does not provide infinite contrast between written and non-written areas. This is because a structure is defined using, for example, a raster scan while modulating the ion beam current. The ion beam cannot be completely turned off when moving sequentially from one pixel to the next, leading to unwanted milling of the nominally unwritten area. However, the contrast can be increased choosing an appropriate milling strategy, as we will show in this paper. Besides the contrast, the scan path of the ion beam can also influence the geometry of the milled structure. The most common scan routines are serpentine, raster, annulus and radial (or spiral), which are described in [35, 43]. The main method for analysing the geometry is cross-sectioning of the holes and subsequent inspection by scanning electron microscopy (SEM). The SEM photos are then further analysed to find the milling depths, sidewall angles and average hole diameters. The exact cross-sectioning method is presented first in detail in section 2.2, because we found that some cross-sectioning methods might influence the originally milled geometry leading to incorrect interpretations. In section 3, we compare the hole geometry obtained for two routines for bulk Si, SOI and Si membranes, namely raster and spiral scanning. The remainder of this paper refers to results obtained by using the spiral-scan routine. To optimize the sidewall angles of the submicrometre holes needed for photonic operation further, we investigated in section 4 the impact of a range of doses on the milling depth and sidewall angle for both individual holes and holes within a triangular-lattice crystal at 48 pA. Furthermore, in section 5 we present the dwell-time–loopnumber-combination variation for a milling current of 48 pA, while keeping the total dose constant. Finally, the main conclusions are presented in section 6.

of the group velocity in PCS structures by several orders of magnitude [8] and their application as a tuneable time delay has also been reported [9]. A typical PCS structure consists of a high-refractive-index layer (e.g. Si, GaAs) perforated with a 2D periodic lattice of air holes having a diameter of approximately half the lattice period. The layer is stacked between two semi-infinite claddings that can be either patterned or homogeneous. There are two coexisting confinement mechanisms of light in PCS: in-plane confinement by the photonic bandgap effect [10] and vertical confinement by refractive-index-guiding realized by an index contrast between the PCS and its claddings. The basic functional PCS-based building blocks (resonators, waveguides, filters, switches, etc) are relatively small structures (10–40 μm), which can be combined to form more complex large-scale architectures. Desired milling depths for photonic operation and moulds are between 100 and 200 nm [11] and the typical hole size is about 200–300 nm in diameter, arranged in a lattice with a periodicity around 0.5 μm [5, 7, 8, 12–14]. Focused ion beam (FIB) milling can be used to define these PCS structures in, for example, silicon, which can be used as a mould for nano-imprint lithography, or in a silicon on insulator (SOI) layer configuration suitable for photonic applications [14]. An important challenge for nanophotonics is its interfacing to the outside world. For guided-wave applications, it is often necessary to use relatively large-sized access waveguides that are usually defined using conventional lithography and etching methods. On the other hand, we have the small photonic crystal elements with features requiring sub-10 nm resolution, which should be accurately aligned with respect to the larger structures. Due to its inherent imaging and high-accuracy alignment properties, FIB is a promising tool for fabricating nanophotonic structures on substrates containing predefined microscopic and macroscopic waveguiding structures. In opto-electronics, FIB has been applied for fabrication of micro-optical components with low surface roughness [15] and for defining the end facet mirrors for conventional semiconductor lasers [16]. It has also been applied for bulk micromachining of macro-porous silicon in order to fabricate 3D Yablonovite-like photonic crystals [17], and for fabricating 2D periodic, metal nanostructures [18] of which the shape of the primitive unit cell can be varied [19]. Another application of FIB has been the fabrication of quasi-2D photonic crystals by milling 1D gratings into freestanding multilayer membranes [20]. Direct FIB milling in silicon is still in an early state of maturity, the losses by modifications of the crystalline Si [21–23] induced by ion bombardment and implantation of, for example, gallium ions within the sample [24] are still too high [25] to realize highquality PCS devices. However, some recent results have been achieved in lowering the losses by either using etchassisting gasses [25, 26] or heat-treatments (out-diffusion of ions) [27, 28], which may lead to breakthroughs on this subject. The damaged Si at the Si–air interface may be removed by wet chemical etching [29]. On the other hand, FIB processing is an ideal candidate for fabrication of moulds, where the previously mentioned disadvantages of the ion milling process do not apply [11, 30, 31]. The smoothening effect of FIB processing [32] can, for example, be exploited for creating nanosmooth moulds and replicas.

2. FIB processing 2.1. Machine properties and general settings A dual-beam FIB machine, the Nova 600 from FEI Company, was used for processing our structures. This machine has the advantage that it contains both an ion beam and an e-beam 2


W C L Hopman et al

(a)

(b)

Figure 1. Different methods for cross-sectioning. (a) Direct milling of a rectangular hole into the Si crystal; details can be found in the main text. (b) Cross-section milling after local metal (Pt) deposition.

surface. The figure shows that redeposited material fills up the holes and possibly modifies the shape of the hole. One of the effects of redeposition is a very low contrast between the original un-milled Si and the milled holes. The shape may be partially resolved by using low acceleration voltages and a high resolution SEM machine to minimize the influence of the observed charging effect. However, a better method, which is used for the Si and SOI milling experiments presented in this paper, is first locally depositing a layer of platinum (Pt) using the FIB and a precursor gas (methylcyclopentadienyl)trimethyl-platinum. Second, a hole is milled with a sloped bottom to avoid long milling times. Finally, a line by line scan (termed cleaning cross-section) is applied at a lower current (28 pA) to establish a high-contrast high-resolution image; see figure 1(b). Since the electron beam and the ion beam are arranged at a fixed angle of 52◦ , SEM photos at an angle can be made of the crosssection without rotating the stage. Small angles can be used for visualizing the cross-sections of deep holes to minimize the size and consequently the milling times of the large hole for cross-sectioning. Since the SEM photos are taken under an angle, the horizontal scale bar cannot be applied to the vertical direction. The length of the image in the vertical direction has to be scaled by 1/ sin (angle). The membrane structures were cross-sectioned by milling a slit into the membrane using a current of 48 pA and fine-polishing using a line by line milling strategy at a current of 9.7 pA. Although redeposition on the sidewall cannot be completely avoided when using this procedure, we observed little difference when comparing the cross-sectioned membrane with the angled top view SEM images; an example is shown in figure 2.

column. This powerful combination allows for both chargeneutralization in the case of milling insulating materials, and non-destructive characterization by SEM. The ion beam is generated from a gallium liquid metal source. The field of view (FoV) of this FIB machine is divided into 4096 × 4096 pixels. The actual size of this grid depends on the magnification set prior to the milling. We used a magnification (in a view mode termed Quad-view) of 5000 (5 kX), resulting in a field of view of about 25 × 25 μm2 , and a minimum inter-pixel distance of 6.2 nm. For the experiments reported in this section the dwell time was predefined without optimizing at a value of 0.1 ms, which in our case resulted in a loop time of 6.0 s. All pixels are sequentially specified by their coordinates and dwell time. The 1/e ion beam spot-size was maintained at approximately 1.5 times the inter-pixel distance. This spot-size is mainly determined by the FIB milling current, which can be set in discrete steps by selecting an aperture size. Although the spot-size grows with the milling current, the lowest current (1.5 pA, spot-size = 7.5 nm) is not the optimum for practical reasons like the total milling-time and the feasible accuracy of focusing of the beam. In previous experiments we found that milling at 48 pA, which gives a spot-size of about 18 nm, provides for this FoV a good balance between milling time and accuracy. Switching between currents would be ideal for larger structures; the boundaries of the hole could then be defined at a low current (small spot-size). However, switching between currents and thus apertures is not an option in the FIB used in this research, because each aperture has a different focus and beam offset, which cannot (in our current machine) be programmed. 2.2. Cross-sectioning In this work, cross-sections need to be realized in order to study how geometries of submicrometre-sized holes in several material systems are affected by the milling parameters. We tried several cross-sectioning techniques, namely milling of a hole with a sloped milling depth, line by line polishing, and combinations of these. In figure 1(a), a combination of both methods is used, first a big hole is milled at a current of 48 pA having a sloped bottom with the largest depth found at the cross-section interface; second a line by line milling strategy is applied at a lower current of 28 pA to achieve a smooth

3. Spiral scanning versus raster scanning The standard scanning method used in most FIB applications is the raster scan, which is schematically shown in figure 3(a). The ion beam scans from one side to the other, while it mills for a specified dwell time at the pixel positions provided in the pattern definition file (here termed the stream file). Since the beam is not switched off between the holes, the scan routine causes interhole milling, which leads to the top surface in the entire FoV being lowered with respect to the surrounding 3


W C L Hopman et al

produce more vertical sidewalls, because locally redeposited material is immediately milled away to a large extent. The method is tested on circular holes for reasons of obvious symmetry; however, the spiral scan method is also expected to yield results with respect to the decrease in interhole milling and the sidewall angle for non-circular holes. The spiral scan method may also be beneficial for holes having sharp corners; however, this has not been investigated in this research. In section 3.1 we will compare the two scan routines, applied to ‘bulk’ Si for two doses. In section 3.2 the same experiment is repeated for SOI, and in section 3.3 the scan routines are evaluated on Si membranes. Figure 2. An example of a well defined FIB-milled planar PCS in a Si membrane (viewed at 52◦ ). The hexagons shown here were rotated by 9◦ with respect to the photonic crystal lattice, in order to obtain an enlarged photonic bandgap [13]. Through the structure a ‘shadow pattern’ can be seen in the bottom silicon due to milling of the ion beam into the substrate. A cross-section (viewed at 35◦ ) of the same structure made by milling a slit into the air-bridge PCS is shown in the inset. The structure was milled at 9.7 pA using a dwell time of 0.01 ms.

(a)

3.1. Silicon For the experiments reported here we mounted a piece of ptype Si 100 on a chuck using silver paint for providing a conducting path between the sample and the grounded holder. The electrical contact when using the paint is much better than for example is obtained using double-sided carbon tape. However, in some cases we noticed that the paint had a non-uniform layer thickness, resulting in specimens that were slightly tilted up to 2◦ , which may lead to asymmetrical holes. We designed two stream files defining 50 holes with 250 nm diameter and 440 nm periodicity. The raster scan file defined 64 689 pixels, whereas the spiral scan file comprised 59900 pixels. The difference, which is caused by the spiral pattern filling the circular hole shape more efficiently, leads to small differences in the dose per hole. The two writing patterns were compared for 12 and 36 repetitions (loops), resulting in milling depths of about 250 and 900 nm, respectively. The ratio between the number of loops and the milling depth is not linear due to the redeposition effect, as we will show in section 4. The spread in milling depth will both reveal the best method for fabricating holes suitable for photonic application (12 loops) and it can show the geometry differences when subjected to a high level of redeposition (36 loops). Figure 4(a) shows the result of a 12-loop raster scan. The overview SEM and the cross-section were taken at an angle of 52◦ . Due to the angle, the scale bar in the vertical direction has to be multiplied by sin(52◦ ) to match the scale bar displayed for the horizontal direction, as mentioned in section 2.2. Because the structure of holes is sunk into the host material, we define two milling depths, shown in the inset of figure 4(a). The first

(b)

Figure 3. (a) Raster scanning. (b) Spiral scanning.

surface. We refer to this effect as the structure being ‘sunk’ into the substrate. A better method for milling arrays of holes is spiral scanning. A schematic drawing of the spiral scan is also shown in figure 3(b). The spiral scan significantly reduces the amount of interhole milling, as will be shown experimentally in the next section. This method is expected to

(a)

(b)

Figure 4. Patterns milled at 48 pA using 12 loops and 0.1 ms dwell time per pixel. All SEM images were taken at a 52◦ angle. The insets show the cross-sections. (a) Raster scan; (b) spiral scan.

4


W C L Hopman et al

(a)

(b)

Figure 5. Patterns milled at 48 pA in Si using 36 loops and 0.1 ms dwell time per pixel. The top view SEM images were taken at a 52◦ angle, and the cross-sections shown in the insets were taken at 45◦ . (a) Raster scan; (b) spiral scan.

Table 1. Comparison of raster and spiral milling in Si, SOI and Si membranes (Si-m).

Scan routine

Pixels Material per hole

Loop-number

Absolute depth, Dose/holea , h abs ± 3% D (pC) (nm)

Raster Spiral Raster Spiral Raster Spiral Raster Spiral Raster

Si Si Si Si SOI SOI Si-m Si-m Si-m

12 12 36 36 12 12 13 13 20

75 69 224 207 75 69 81 75 115

a

1293 1198 1293 1198 1293 1198 1293 1198 1293

253 300 906 917 364 355 — — —

Relative depth, h rel ± 3% (nm)

Average diameter, davg ± 5% (nm)

Sidewall angle ±10% (deg)

214 282 855 885 345 355 338 339 470

299 276 249 232 330 262 350 380 366

19 13 9 8 11 7 5 5 2.8 nA), which may explain their findings. The data from which the graphs in figure 11 have been constructed can be found in table 6. Similar experiments were conducted on SOI for two currents; the results are shown in table 2. The values show a close resemblance to the values obtained from the Si milling experiment at 48 pA. Based on these results, we conclude that lowering the current does not lead to a significant reduction in sidewall angle at a dose around 60 pC. This seems to contradict 8


W C L Hopman et al

for currents in the studied ion current range between 9.7 and 48 pA, that there is no direct relation between the current (which determines the spot size and shape of the tails of the beam) and the sidewall angle. The main reason for this is that redeposition is the dominant effect.

Table 2. Loop number variation of a hole within an array in SOI.

Dwell time Current I (pA) Loops (ms)

Relative depth, Dose h rel ± per hole 3% (pC) (nm)



48 9.7 9.7 9.7

61 59 66 66

282 258 255 210

7 7 11 20

2 6 62 617

1 1 0.1 0.01

286 320 349 301

6. Conclusions We have shown that it is necessary to deposit a metal (here platinum) before making cross-sections to avoid perturbation of submicrometre milled holes through redeposition, and to achieve a large contrast. The large contrast was needed to determine parameters such as the sidewall angle, milling depth and hole diameter with high precision. Two different routines for scanning the beam over the sample were compared for milling in silicon, SOI and silicon membranes: raster scanning and spiral scanning. The spiral method has the advantage that the amount of structure lowering (sinking of the structure within the substrate) and interhole milling can be reduced significantly. We also found a slight increase in verticality of the sidewall angles and an increase in symmetry of the hole for the spiral scan routine applied to silicon and SOI. The study on the silicon membranes showed that, using the spiral method, it is possible to achieve sidewall angles as small as 1.5◦ , which is comparable to state of the art dry etching achievements in silicon. Further optimizations were explored by varying the dose applied to a single hole and to a hole as part of an array, while fixing the dwell time at 0.1 ms. We found that the sidewall angle at half the hole depth can be decreased to values of about 5◦ in Si. Furthermore, we experimentally showed the non-linear relationship between the depth and dose for submicrometre holes. We found that due to the proximity effects presented here, we obtain a slightly higher milling rate for the holes in an array than for single holes. In the studied dose range we observed also a constant difference between the absolute depth h abs and the relative depth h rel . The impact of the combination dwell time and loop number was investigated by fixing the dose at 69 pC and varying the number of loops between two and 200 and the dwell time between 0.6 ms and 6 μs in Si. The most favourable combination with respect to the hole geometry was found to be a small loop number in combination with a relatively high dwell time of 1 ms. The results seem to contradict with other findings in which smaller dwell times are recommended. Therefore, we assume that the number of loops at a fixed dose per hole is the best optimization

Table 3. Values obtained from milling a single 250 nm diameter hole in Si.

Loops

Dose per holea , D (pC)

Absolute depth, h abs ± 3% (nm)



3 6 12 30 60 90 120 200

17 35 69 173 345 518 690 1150

84 169 328 683 1043 1305 1481 1814

300 285 210 150 178 240 250 273

50 29 14 9 5 4 4 3

a

Milled at 48 pA using a 0.1 ms dwell time.

Table 4. Values obtained from milling a single 300 nm diameter hole in Si. Milling time, T (s)

Dose/ holea , D (pC)



Sidewall angle ϕ ± 10% (deg)

1 2 5 10 15 25

48 96 240 480 720 1200

191 380 1007 1428 1624 1823

349 322 280 283 305 291

27 9 8 6 5 4

a

Milled at 48 pA using a 0.1 μs dwell time.

the observation published in [48], where the sidewall angle is related to the width of the Gaussian beam shape. However, the difference may be explained by the fact that the authors also modify the dose per pixel when varying the current and the investigated dose and hole diameter differ. Our results suggest,

Table 5. Values obtained from milling an arraya of holes in silicon.

Loop number

Dose per holeb D (pC)





3 6 12 30 60 90

17 35 69 173 345 518

83 160 320 872 1168 1312

68 141 292 766 1065 1224

310 290 261 220 222 217

53 20 11 8 6 5

a b

Lattice constant 440 nm, hole diameter 250 nm. Milled at 48 pA using a 0.1 ms dwell time.

9


W C L Hopman et al

Table 6. Effect of dosea –loop number variation in an arrayb of holes.

Loops

Dwell time (ms)





2 6 20 120 200

0.6 0.2 0.06 0.001 0.006

396 355 345 316 314

386 345 335 307 294

252 266 281 287 281

8.4 10.0 11.8 12.2 13.4

a b

The dose was fixed at 69 pC. Lattice constant 440 nm, hole diameter 250 nm.

parameter with respect to the sidewall angles (redeposition) and not necessarily only the dwell time. In SOI we found a similar result for the combination of dwell time and loop number for beam currents of 9.7 and 48 pA.

Acknowledgments This research was supported by NanoNed, a national nanotechnology program coordinated by the Dutch ministry of Economic Affairs, and was also supported by the European Network of Excellence ePIXnet. Laurens Kuipers would like to acknowledge the ‘Stichting voor Fundamenteel Onderzoek der Materie (FOM)’, which is financially supported by the ‘Nederlandse Organisatie voor Wetenschappelijk Onderzoek (NWO)’.

References [1] Melloni A, Monguzzi P, Costa R and Martinelli M 2003 Design of curved waveguides: the matched bend J. Opt. Soc. Am. A 20 130–37 [2] Villeneuve P R, Fan S and Joannopoulos J 1996 Microcavities in photonic crystals: mode symmetry, tunability, and coupling efficiency Phys. Rev. B 54 7837–42 [3] Chakravarty S, Topol’ancik J, Bhattacharya P, Chakrabarti S, Kang Y and Meyerhoff M E 2005 Ion detection with photonic crystal microcavities Opt. Lett. 30 2578–80 [4] Noda S, Yokoyama M, Imada M, Chutinan A and Mochizuki M 2001 Polarization mode control of two-dimensional photonic crystal laser by unit cell structure design Science 293 1123–5 [5] Akahane Y, Asano T, Song B S and Noda S 2003 High-Q photonic nanocavity in a two-dimensional photonic crystal Nature 425 944–7 [6] Kuramochi E, Notomi M, Mitsugi S, Shinya A, Tanabe T and Watanabe T 2006 Ultrahigh-Q photonic crystal nanocavities realized by the local width modulation of a line defect Appl. Phys. Lett. 88 041112 [7] Notomi M, Shinya A, Mitsugi S, Kira G, Kuramochi E and Tanabe T 2005 Optical bistable switching action of Si high-Q photonic-crystal nanocavities Opt. Express 13 2678–87 [8] Gersen H, Karle T J, Engelen R J P, Bogaerts W, Korterik J P, van Hulst N F, Krauss T F and Kuipers L 2005 Direct observation of Bloch harmonics and negative phase velocity in photonic crystal waveguides Phys. Rev. Lett. 94 123901 [9] Vlasov Y A, O’Boyle M, Hamann H F and McNab S J 2005 Active control of slow light on a chip with photonic crystal waveguides Nature 438 65–9 [10] Johnson S G, Fan S H, Villeneuve P R, Joannopoulos J D and Kolodziejski L A 1999 Guided modes in photonic crystal slabs Phys. Rev. B 60 5751–8

10

[11] Chou S Y, Krauss P R and Renstrom P J 1996 Nanoimprint lithography J. Vac. Sci. Technol. B 14 4129–33 [12] Hopman W C L, Van Der Werf K O, Hollink A J F, Bogaerts W, Subramaniam V and De Ridder R M 2006 Nano-mechanical tuning and imaging of a photonic crystal micro-cavity resonance Opt. Express 14 8745–52 [13] Bostan C G and de Ridder R M 2002 Design of photonic crystal slab structures with absolute gaps in guided modes J. Optoelectron. Adv. Mater. 4 921–8 [14] Bostan C G, De Ridder R M, Gadgil V J, Kelderman H, Kuipers L and Driessen A 2004 Design and fabrication of line-defect waveguides in hexagon-type SOI photonic crystal slabs: Photonics Europe (Strasbourg) Proc. SPIE 5450 323–32 [15] Fu Y and Bryan N K A 2001 Experimental study of microcylindrical lenses fabricated using focused-ion-beam technology J. Vac. Sci. Technol. B 19 1259–63 [16] Ito T, Ishikawa H, Egawa T, Jimbo T and Umeno M 1997 Fabrication of flat end mirror etched by focused ion beam for GaN-based blue-green laser diode Japan. J. Appl. Phys. 36 7710–1 [17] Chelnokov A, Wang K, Rowson S, Garoche P and Lourtioz J M 2000 Near-infrared Yablonovite-like photonic crystals by focused-ion-beam etching of macroporous silicon Appl. Phys. Lett. 77 2943–5 [18] Lezec H J and Thio T 2004 Diffracted evanescent wave model for enhanced and suppressed optical transmission through subwavelength hole arrays Opt. Express 12 3629–51 [19] Klein Koerkamp K J K, Enoch S, Segerink F B, van Hulst N F and Kuipers L 2004 Strong influence of hole shape on extraordinary transmission through periodic arrays of subwavelength holes Phys. Rev. Lett. 92 183901 [20] Wang K, Filloux P, Paraire N, Roca I and CabarrocasBulkin P 2003 Two-dimensional photonic crystals by focused-ion-beam etching of multilayer membranes J. Vac. Sci. Technol. B 21 966–9 [21] Brezna W, Wanzenböck H, Lugstein A, Bertagnolli E, Gornik E and Smoliner J 2003 Scanning capacitance microscopy investigations of focused ion beam damage in silicon Physica E 19 178–82 [22] Pellerin J G, Shedd G M, Griffis D P and Russell P E 1989 Characterization of focused ion-beam micromachined features J. Vac. Sci. Technol. B 7 1810–2 [23] Kim Y K, Danner A J, Raftery J J and Choquette K D 2005 Focused ion beam nanopatterning for optoelectronic device fabrication IEEE J. Sel. Top. Quantum Electron. 11 1292–8 [24] Knights A P and Hopper G E 2003 Effect of ion implantation induced defects on optical attenuation in silicon waveguides Electron. Lett. 39 1648–9 [25] Cryan M J, Hill M, Sanz D C, Ivanov P S, Heard P J, Tian L, Yu S Y and Rorison J M 2005 Focused ion beam-based fabrication of nanostructured photonic devices IEEE J. Sel. Top. Quantum Electron. 11 1266–77 [26] Schrauwen J, Van Thourhout D and Baets R 2006 Focused-ion-beam fabricated vertical fiber couplers on silicon-on-insulator waveguides Appl. Phys. Lett. 89 141102


W C L Hopman et al

[27] Chyr I, Lee B, Chao L C and Steckl A J 1999 Damage generation and removal in the Ga+focused ion beam micrmachining of GaN for photonic applications J. Vac. Sci. Technol. B 17 3063–7 [28] Tanaka Y, Tymczenko M, Asano T and Noda S 2006 Fabrication of two-dimensional photonic crystal slab point-defect cavity employing local three-dimensional structures Japan. J. Appl. Phys. 1 45 6096–102 [29] Lehrer C, Frey L, Petersen S, Ryssel H, Schafer M and Sulzbach T 2004 Integration of field emitters into scanning probe microscopy sensors using focused ion and electron beams J. Vac. Sci. Technol. B 22 1402–6 [30] Chaix N, Landis S, Hermelin D, Leveder T, Perret C, Delaye V and Gourgon C 2006 Influence of mold depth on capillary bridges in nanoimprint lithography J. Vac. Sci. Technol. B 24 3011–5 [31] Li H W, Kang D J, Blamire M G and Huck W T S 2003 Focused ion beam fabrication of silicon print masters Nanotechnology 14 220–3 [32] Hopman W C L, De Ridder R M, Selvaraja S, Bostan C G, Gadgil V J, Kuipers L and Driessen A 2006 Realization of 2-dimensional air-bridge silicon photonic crystals by focused ion beam, milling and nanopolishing: photonics Europe (Strasbourg) Proc. SPIE 6182 167–73 [33] Tanaka Y, Asano T, Akahane Y, Song B S and Noda S 2003 Theoretical investigation of a two-dimensional photonic crystal slab with truncated cone air holes Appl. Phys. Lett. 82 1661–3 [34] Prenitzer B I, Urbanik-Shannon C A, Giannuzzi L A, Brown S R, Irwin R B, Shofner T L and Stevie F A 2003 The correlation between ion beam/material interactions and practical FIB specimen preparation Microsc. Microanal. 9 216–36 [35] Fu Y, Bryan N K A, Shing O N and Hung N P 2000 Influence of the redeposition effect for focused ion beam 3D micromachining in silicon Int. J. Adv. Manuf. Technol. 16 877–80 [36] Frey L, Lehrer C and Ryssel H 2003 Nanoscale effects in focused ion beam processing Appl. Phys. A 76 1017–23

[37] Lipp S, Frey L, Lehrer C, Frank B, Demm E and Ryssel H 1996 Investigations on the topology of structures milled and etched by focused ion beams J. Vac. Sci. Technol. B 14 3996–9 [38] Balasubramanian K, Heard P J and Cryan M J 2006 Focused ion beam fabrication of two dimensional photonic crystals in silicon-on-insulator J. Vac. Sci. Technol. B 24 2533–7 [39] Yamaguchi H, Shimase A, Haraichi S and Miyauchi T 1984 Characteristics of silicon removal by fine focused gallium ion beam J. Vac. Sci. Technol. B 3 71–4 [40] Fu Y, Bryan N K A, Shing O N and Wyan H N P 2000 Influence analysis of dwell time on focused ion beam micromachining in silicon Sensors Actuators A 79 230–4 [41] Adams D P and Vasile M J 2006 Accurate focused ion beam sculpting of silicon using a variable pixel dwell time approach J. Vac. Sci. Technol. B 24 836–44 [42] Nellen P M, Callegari V and Brönnimann R 2006 FIB-milling of photonic structures and sputtering simulation Microelectron. Eng. 83 1805–8 [43] Tseng A A 2004 Recent developments in micromilling using focused ion beam technology J. Micromech. Microeng. 14 R15–34 [44] Lugstein A, Basnar B, Smoliner J and Bertagnolli E 2003 FIB processing of silicon in the nanoscale regime Appl. Phys. A 76 545–8 [45] Bischoff L, Teichert J and Heera V 2001 Focused ion beam sputtering investigations on SiC Appl. Surf. Sci. 184 372–6 [46] McNab S J, Moll N and Vlasov Y A 2003 Ultra-low loss photonic integrated circuit with membrane-type photonic crystal waveguides Opt. Express 11 2927–39 [47] Settle M, Salib M, Michaeli A and Krauss T F 2006 Low loss silicon on insulator photonic crystal waveguides made by 193 nm optical lithography Opt. Express 14 2440–5 [48] Lehrer C, Frey L, Petersen S and Ryssel H 2001 Limitations of focused ion beam nanomachining J. Vac. Sci. Technol. B 19 2533–8 [49] Wilhelmi O, Reyntjens S, Wall D, Geurts R, Jiao C and Roussel L 2006 Nanofabrication and Rapid Prototyping with DualBeam Instruments: Conf. Micro- and Nano-Engineering (Barcelona)

11